Investigating oral performance on controlled versus spontaneous speaking tasks: Implications for second language pedagogy and assessment

Assessment tasks of second language (L2) proficiency can be loosely classified into controlled and spontaneous tasks. While spontaneous tasks are widely accepted as authentic proficiency instruments, they can be time-consuming and labor-intensive to develop, administer, and score. In contrast, controlled tasks are more flexible and efficient, and have gained popularity with recent technological innovations. Although research has shown that controlled tasks are highly predictive of spontaneous performance, the question remains as to whether the two types of tasks measure similar kinds of language use (i.e., construct). To address this concern, this study examines characteristics of performance regarding both the product and process of language use on controlled versus spontaneous tasks to elucidate the constructs measured by these two types of tasks. This study focuses on speaking proficiency by examining elicited imitation (EI) and the Oral Proficiency Interview (OPI), with each representing controlled and spontaneous speaking tasks, respectively. Employing a mixed-methods design, this study compared speaking performances of 75 L2 learners and 20 native speakers of Chinese on an EI and an advisory simulated OPI test. Test-taking processes and strategies were elicited through the use of questionnaires administered to a subgroup of 16 L2 learners and eight native speakers. Test responses were quantitatively analyzed for fluency, accuracy, and complexity features, and processes and strategies during task completion were qualitatively examined. Results of the quantitative analyses showed that speed fluency, accuracy, and complexity were elicited similarly by the two tasks, but breakdown and re- pair fluency were elicited differently. Results from the qualitative analyses demonstrated that both tasks required the processes of speech production and L2 learners employed common test-taking strategies to complete the tasks. EI elicited more cognitive strategies to process auditory input, while OPI elicited more metacognitive strategies for planning and communication strategies to deal with content breakdowns. In addition, the results revealed that both tasks tap into core components of language proficiency. The difference lies in that EI disfavors micro-level (dis)fluency, while OPI tolerates discourse-level (dis)fluency. These findings suggest that both EI and OPI tasks are valid measures of L2 oral proficiency, as reflected by their similarities in language use, but they also have unique task features, as revealed by the differences in language use. These findings have implications regarding the construct validity of controlled and spontaneous speaking tasks and can guide teaching and assessment practices in L2 Chinese as well as in other languages.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

E Asian Languages & Cultures

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2022

Author dc:creator

• Lei, Yuyun

Contributors dc:contributor

• Yan, Xun
• Shih, Chilin
• Bowles, Melissa A.
• Packard, Jerome L.

• Language assessment
• Elicited imitation
• Oral proficiency interview
• Second Language Proficiency
• Mandarin Chinese

Statement dc:rights

• Copyright 2021 Yuyun Lei

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/113154

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/113154